Surface enhanced Raman spectroscopy on a flat graphene surface Weigao Xua, Xi Linga, Jiaqi Xiaoa, Mildred S. Dresselhausb,c, Jing Kongb, Hongxing Xud, Zhongfan Liua, and Jin Zhanga,1 aCenter for Nanochemistry, Beijing National Laboratory for Molecular Sciences, Key Laboratory for the Physics and Chemistry of Nanodevices, State Key Laboratory for Structural Chemistry of Unstable and Stable Species, College of Chemistry and Molecular Engineering, Peking University, Beijing 100871, China; bDepartment of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA 02139; dBeijing National Laboratory of Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China; and cDepartment of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139 Edited by Nicholas J. Turro, Columbia University, New York, NY, and approved April 18, 2012 (received for review April 3, 2012) Surface enhanced Raman spectroscopy (SERS) is an attractive give rise to greatly enhanced local electromagnetic fields via analytical technique, which enables single-molecule sensitive de- the localized surface plasmon resonance effect. The localized tection and provides its special chemical fingerprints. During the electromagnetic field (usually called a “hot” spot) thus results in past decades, researchers have made great efforts towards an ideal the dominant contribution of SERS enhancement [referred to as SERS substrate, mainly including pioneering works on the prepara- electromagnetic enhancement (EM)], which can reach 108 or tion of uniform metal nanostructure arrays by various nanoassem- more (8–10). Usually there is another chemical contribution bly and nanotailoring methods, which give better uniformity and called chemical enhancement (CM), but with a minor enhance- reproducibility. Recently, nanoparticles coated with an inert shell ment factor (typically 10 to 102) (11, 12). As a result, molecules were used to make the enhanced Raman signals cleaner. By depos- adsorbed near the electromagnetic hot spots dominate the Ra- iting SERS-active metal nanoislands on an atomically flat graphene man intensity. The distribution of molecules on a normal SERS layer, here we designed a new kind of SERS substrate referred to as substrate is usually complicated, as the molecules near the hot a graphene-mediated SERS (G-SERS) substrate. In the graphene/ spots can be in fluctuating amounts and random orientations. metal combined structure, the electromagnetic “hot” spots (which On the other hand, chemical interactions between the molecules is the origin of a huge SERS enhancement) created by the gapped and the metal substrate can make this case more complicated. CHEMISTRY metal nanoislands through the localized surface plasmon resonance Chemical adsorption-induced vibrations, molecular deformation effect are supposed to pass through the monolayer graphene, re- and distortion, charge transfer between the metal and molecules, sulting in an atomically flat hot surface for Raman enhancement. photocarbonization, photobleaching or metal-catalyzed side re- Signals from a G-SERS substrate were also demonstrated to have actions (10, 13) may all affect the final signal and make it difficult interesting advantages over normal SERS, in terms of cleaner vibra- to precisely assign each vibration of a SERS spectrum. Thus, tional information free from various metal-molecule interactions the SERS signal is sometimes too sensitive to a SERS substrate, and being more stable against photo-induced damage, but with possible fluctuations caused by nonuniform molecular adsorption a comparable enhancement factor. Furthermore, we demonstrate (amounts/configurations), different metal enhancers and the the use of a freestanding, transparent and flexible “G-SERS tape” corresponding metal-molecule interactions bring down the com- (consisting of a polymer-layer-supported monolayer graphene with parability between parallel SERS experiments, and as a result sandwiched metal nanoislands) to enable direct, real time and reli- standard analytical methods based on SERS have been rarely able detection of trace amounts of analytes in various systems, popularized. To further understand the SERS effect and fully which imparts high efficiency and universality of analyses with excite its advanced applications, a SERS substrate with a more G-SERS substrates. “intelligent” form is still lacking. From the above, the availability of an enhancement substrate, atomically smooth substrate ∣ metal-molecule isolation ∣ signal fluctuation ∣ which can give a uniform, stable, clean and highly sensitive SERS mediator ∣ application response has remained a bottleneck for extending the further applications of SERS (10, 14). Current efforts towards more or almost all sorts of analytical methods, the dream to improve reproducible SERS analyses are mainly focused on the nanotai- Ftheir sensitivity as well as their reproducibility and to optimize loring or nanoassembly of substrates with uniform metal nanos- the analytical process (e.g., to simplify the sample preparation/ tructure arrays (15–19). However, a “uniform” SERS substrate is measurement procedures for quick analysis and to enable in situ still not uniform on the nanoscale, the distribution of molecules and real time monitoring) is a long-term pursuit. Spectroscopic and their states may be quite uncontrollable in the vicinity of the approaches based on fluorescence, infrared absorption and electromagnetic hot spots. On the other hand, metal nanoparti- Raman scattering have been developed with rising importance cles with a thin and inert SiO2 shell were recently demonstrated for various sensing and imaging applications. Among them, to make SERS signals cleaner (13). Yet, the coating of a pinhole- Raman scattering provides more structural information (charac- free and thin SiO2 layer (to avoid metal-molecule interaction teristic vibrational information of each chemical bond) over without large sacrifice to the SERS-activity) is an over delicate fluorescence, and higher spatial resolution (shorter excitation technique. In this work, based on the use of the two-dimensional wavelength) over infrared absorption (1). However, as Raman atomic crystal graphene, we designed a unique kind of SERS scattering is an inelastic scattering process with a very low cross- section, it is not very sensitive and thus has limited analysis effi- Author contributions: W.X., X.L., and J.Z. designed research; W.X., X.L., J.X., and H.X. ciency and applicability. performed research; H.X. and J.Z. contributed new reagents/analytic tools; W.X., X.L., J.X., Because of this, surface enhanced Raman spectroscopy M.S.D., J.K., Z.L., and J.Z. analyzed data; and W.X., X.L., J.X., M.S.D., J.K., and J.Z. wrote the (SERS) has been developed since the 1970s (2–4) to enable ul- paper. trasensitive characterization down to the single-molecule level (5, The authors declare no conflict of interest. 6), comparable to single-molecule fluorescence spectroscopy (7). This article is a PNAS Direct Submission. In a SERS experiment, a rough metal surface or colloidal metal 1To whom correspondence should be addressed. E-mail: [email protected]. nanostructures are typically used. Here, the highly curved or This article contains supporting information online at www.pnas.org/lookup/suppl/ gapped metal regions (under proper incident light conditions) doi:10.1073/pnas.1205478109/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1205478109 PNAS Early Edition ∣ 1of6 Downloaded by guest on September 29, 2021 A monolayer graphene D (1LG) metal nanoislands Normal SERS supporting layer G-SERS tape strong B C E 0.5 nm SERS (G-SERS) 1LG 1LG 1LG Graphene-mediated Au Electromagnetic field Au/Ag weak Fig. 1. Design of a G-SERS substrate. (A) Components of a G-SERS tape. (B) The simulated electromagnetic field distribution of a 60-nm gold hemisphere dimer covered with graphene (with a 2-nm gap between two gold hemispheres and a 0.5-nm gap between gold hemispheres and graphene). The excitation wavelength is 632.8 nm, with the incident direction perpendicular to the graphene surface (from top to bottom). (C) The resulting “hot” graphene surface to serve as a G-SERS substrate. (D, E) Schematic illustration of molecules adsorbed on a normal SERS substrate (D) and on a G-SERS substrate (E). substrate containing a graphene/metal-combined structure, re- in a more controllable way, (ii) a spacer to separate the metal- ferred to as a graphene-mediated SERS (G-SERS) substrate. molecule contact, (iii) additional effects like a stabilizer of both the substrate and the molecules under laser exposure. Results To seek the possible superiorities of a G-SERS substrate, a As illustrated in Fig. 1, the active surface of our G-SERS rational way was used to fabricate substrates with both G-SERS substrate is a monolayer graphene (1LG), with gold/silver nanois- regions and normal SERS regions for comparison. As shown in lands tightly adhered on the backside. By collecting electromag- Fig. 2A, a uniform layer of molecules, such as rhodamine 6G netic hot spots from the metal nanoislands on a graphene surface, (R6G) or copper phthalocyanine (CuPc), were first deposited a G-SERS substrate is anticipated to offer an atomically smooth on a SiO2∕Si substrate by vacuum thermal evaporation. Mechani- surface for controllable molecular arrangements as well as well- cally exfoliated graphene pieces were then transferred to the top defined molecular interactions, yet also has the electromagnetic